Potential-Driven Mobility Partitioning for Next-Generation Networks

Kuruc, Gabor

Heszberger, Zalan

Biro, Jozsef

2026-05-27T09:22:21Z

2026

Mobile networks face increasing challenges in managing user mobility, particularly in minimizing frequent handovers that generate signaling overhead and degrade service continuity. This paper develops a mobility modeling framework that combines attraction-repulsion potential fields with softmax-based maximum entropy transitions, specifically tailored for mobile network management and domain formation. In the proposed model, user movement tendencies are influenced by spatial features such as points of interest and repulsive zones, giving rise to a continuous and interpretable potential field over the cell topology. Transition probabilities are then derived via a maximum entropy formulation, ensuring a balance between stochastic mobility and attraction-driven behavior. Building on this foundation, we address the problem of mobility domain formation—partitioning the cellular network into regions that minimize inter-domain transitions. We formulate this as a weighted graph partitioning task based on softmax transition probabilities and show that the structure of our model enables efficient approximate solutions using spectral or modularity-based methods. Compared to traditional empirical approaches relying on raw handover statistics, our model provides smoother, more stable, and semantically meaningful mobility estimates. While maximum entropy and gravity-based mobility models have been applied in broader spatial interaction contexts, to the best of our knowledge, this is the first work to integrate potential field mobility modeling with entropy-based decision dynamics for the specific goal of domain optimization in mobile networks. This framework offers a promising direction for scalable, interpretable, and data-efficient mobility-aware network slicing and management.

angol

Könyvfejezet

application/pdf

http://hdl.handle.net/10890/64943